Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Xtgls: model fitness test


From   yanfenz@CLEMSON.EDU
To   statalist@hsphsun2.harvard.edu
Subject   st: Xtgls: model fitness test
Date   Tue, 16 Sep 2008 18:07:54 -0400 (EDT)

Hi, all:

        I am doing a study examining the impact of relationships among
founders on new bank performance. After running STATA with xtgls,
I realized that there is no way to conduct model fitness test.
After checking STATA FAQ, I learned that there is no R-sq
statistic for xtgls. Then what is the appropriate way to test
whether adding one or more variables increases model fitness?  It
appears that log likelihood test (LRT) is not working because log
likelihood numbers produced by xtgls do not  monotonically
increase with inclusion of more variables. Any suggestion or
explanation? Thanks for your help.

------------------------------------------
Yanfeng Zheng
Assistant Professor
117C Sirrine Hall
Dept. of Management
Clemson University
Clemson, SC 29634
(o)864-656-1573
e-mail: yanfenz@clemson.edu
-------------------------------------------

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index